Search results

[cuDNN][SDPA] Upgrade to cuDNN 9.13.1 patch release

ciflow/binariesTrigger all binary build and upload jobs on the PRTrigger all binary build and upload jobs on the PR

module: cudnnRelated to torch.backends.cudnn, and CuDNN supportRelated to torch.backends.cudnn, and CuDNN support

module: sdpaAll things related to torch.nn.functional.scaled_dot_product_attentiionAll things related to torch.nn.functional.scaled_dot_product_attentiion

open source

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 164033

#164033 In pytorch/pytorch;

· eqy opened on Sep 27, 2025

[CUDA][Memory Snapshot] <code>test_memory_compile_regions</code> appears to interfere with subsequent <code>record_memory_history</code> tests

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: testsIssues related to tests (not the torch.testing module)Issues related to tests (not the torch.testing module)

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#163202 In pytorch/pytorch;

· eqy opened on Sep 17, 2025

[CUDA] hack for host-memory stats test

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

open source

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 163119

#163119 In pytorch/pytorch;

· eqy opened on Sep 16, 2025

[CUDA13][CUDA Graphs] Fix <code>test_graph_external_wait_and_record</code> to account for minimum supported compute-capability

ciflow/trunkTrigger trunk jobs on your pull requestTrigger trunk jobs on your pull request

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: cuda graphsAbility to capture and then replay streams of CUDA kernelsAbility to capture and then replay streams of CUDA kernels

open source

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 163001

#163001 In pytorch/pytorch;

· eqy opened on Sep 15, 2025

[FP8][H100][TF32] Disable tf32 for emulated reference computation in <code>test_scaled_mm_vs_emulated_block_wise</code>

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: floatx (formerly float8)For torch.float8_e5m2 and torch.float8_e4m3 and other sub 8-bit float typesFor torch.float8_e5m2 and torch.float8_e4m3 and other sub 8-bit float types

module: tf32Related to tf32 data formatRelated to tf32 data format

open source

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 162997

#162997 In pytorch/pytorch;

· eqy opened on Sep 15, 2025

[B200][grouped gemm] <code>test_grouped_gemm_compiled_op</code> fails on repeat

matrix multiplication

module: inductor

oncall: pt2

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#162970 In pytorch/pytorch;

· eqy opened on Sep 15, 2025

[DO NOT MERGE][B200] remove xfail of <code>grouped_gemm_compiled_op</code> in <code>test_matmul_cuda.py</code>

BlackwellSpecific failures or issues related to sm100 + Cuda archesSpecific failures or issues related to sm100 + Cuda arches

matrix multiplication

open source

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 162641

#162641 In pytorch/pytorch;

· eqy opened on Sep 10, 2025

[CUDA][cuBLAS][TF32] Skip checking TF32 options and setting cuBLAS handle TF32 modes for other dtypes

module: cublasProblem related to cublas supportProblem related to cublas support

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: tf32Related to tf32 data formatRelated to tf32 data format

open source

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 161823

#161823 In pytorch/pytorch;

· eqy opened on Aug 29, 2025

[CUDA] Don't import hipify modules on CUDA

ciflow/inductor

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: inductor

open source

release notes: inductor (aoti)

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 161107

#161107 In pytorch/pytorch;

· eqy opened on Aug 20, 2025

<code>test_non_contiguous_input_mm_plus_mm</code> looks broken on B200

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: tf32Related to tf32 data formatRelated to tf32 data format

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#159914 In pytorch/pytorch;

· eqy opened on Aug 5, 2025

[CUDA] Add <code>cuda</code> to the names of tests that use CUDA in <code>test_nn.py</code> also some misc. cleanups

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

open source

Stale

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 159600

#159600 In pytorch/pytorch;

· eqy opened on Jul 31, 2025

[cuDNN][CUDA] Allow Flash attention or cuDNN attention as default in fused_sdp_choice test

ciflow/trunkTrigger trunk jobs on your pull requestTrigger trunk jobs on your pull request

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: cudnnRelated to torch.backends.cudnn, and CuDNN supportRelated to torch.backends.cudnn, and CuDNN support

module: sdpaAll things related to torch.nn.functional.scaled_dot_product_attentiionAll things related to torch.nn.functional.scaled_dot_product_attentiion

open source

Stale

topic: not user facingtopic categorytopic category

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open (in progress).

pytorch/pytorchnumber 159582

#159582 In pytorch/pytorch;

· eqy opened on Jul 31, 2025